This research has contributed to an improved understanding of near-bed flow in rivers and the advancement of modelling flow and sediment transport in multiple dimensions. The understanding and prediction of the hydraulic behaviour of river channels are essential to water resources development and river-engineering activity planning. River flow features turbulence and complicated velocity distribution, especially near the bed. An ice cover typically presents in northern rivers during the winter and influences the flow underneath; this further complicates the velocity distribution. The problem of flow near the boundaries (the bed and ice) is notoriously difficult to tackle because of strong velocity shear, multiple length and associated time-scale motions and bed sediment movement. In spite of previous research efforts focusing on the problem, many issues are still unresolved. 
This study has resolved the issue with respect to the link between near-bed flow and flow-induced bed shear stress in a computationally efficient manner. The research work consists of (a) derivation of hydraulic parameters necessary for describing and modelling the velocity field and (b) prediction of the bed shear stress τb and resultant sediment transport along the riverbed (bedload). In part (a), a large volume of winter observations of water velocity from ice-covered Canadian rivers have been obtained. Assume that the velocity distribution between the bed and ice can be described as a two-layer system. Multi-parameter regression analyses are performed on the observations, yielding a function with two exponents and one coefficient as hydraulic parameters. These parameters reveal the relative importance of the bed and ice’s influence on velocity distribution. The function describes the vertical distribution of velocity. Its practical significance includes convenient estimates of winter discharge, which is expensive and extremely difficult to measure in the field. The observations have also been analysed to produce energy and momentum coefficients. These coefficients are rarely available but are necessary input to one-dimensional flow predictions. Additionally, part (a) includes the development of a mathematical model based on the boundary layer theory and application of it to the bed-influenced layer of ice-covered river flow for determining the drag coefficient. The concept of drag coefficient is widely used to give dynamic condition at the riverbed for predicting flow in three dimensions. Part (b) deals with the key issue of τb for bedload computations. An existent multi-layer hydrodynamics model has been extended to explore methods useful to link τb to near-bed flow. Such a link will improve computational efficiency. The model is applied to flow over gravel river dunes – a case of complicated velocity distributions. The model results of velocity and τb are shown to agree well with acoustic Doppler velocimeter measurements from flume experiments. The predicted τb values are used to compute fractional transport rates of non-uniform sediments over the dunes. For bedload modelling, the logarithmic law is shown to provide an appropriate link between near-bed flow and τb; this law should be applied to velocities at a wall distance of approximately 300. When using the multi-layer modelling approach, one should allow a minimum of five layers to resolve the velocity structure from the bed to the wall distance.